Generosity among Friends: Population Homogeneity,
Altruism and Insurance as Determinants of
Raymond M. Duch
University of Oxford
0X1 1NF Oxford UK
University of Oxford
0X1 1NF Oxford UK
June 12, 2014
Altruism is an important omitted variable in much of the Political Economy
literature. While material self-interest is the base of most approaches to redistri-
bution (ﬁrst aﬀecting preferences and then politics and policy), there is a paucity
of research on inequality aversion. We propose (1) that other-regarding concerns
inﬂuence redistribution preferences and (2) that they are conditional on the identity
of the poor. We will show that group homogeneity magniﬁes (or limits) the im-
portance of altruism. In making these distinctions between the poor and the rich,
the arguments in this paper challenge some inﬂuential approaches to the politics of
The considerable scholarship on preferences for redistribution in the population has
not produced a concensus on the micro-foundations of these attitudes. Do the poor want
more redistribution? Its not entirely clear this is the case. Nor does the data necessarily
suggest the rich are strong opponents of redistribution.
Hence this essay has two overriding goals. One of these is simply exploring alternative
strategies for recovering underlying preferences or attitudes. And a second goal is to see
whether more rigorous measurement strategies provide a clearer understanding of these
micro-foundations of attitudes regarding redistribution.
2 The Argument
This paper’s analysis attempts to integrate three distinct approaches to the forma-
tion of preferences for redistribution. The ﬁrst one relies on the idea that the level of
redistribution preferred by a given individual is fundamentally a function of her ma-
terial self-interest (two diﬀerent facets of this argument should be distinguished, one
dealing with redistribution and the other with insurance, risk and mobility). The second
approach maintains that other-regarding concerns matter. Altruistic individuals derive
utility not only from their own material gains but also from those of other people. The
third approach emphasizes identity and in-group solidarity, arguing that ethnic, national
or religious fractionalization reduces overall support for redistribution.
This paper will integrate insights from these three approaches into one argument and
focus on the relationship between in-group identity and altruism. In the following pages,
we will explore in more detail these general frameworks and elucidate this paper’s claims.
In essence, we argue for the importance of non-material factors and propose that they
are conditional on the identity of the poor. Controlling for relative income, we will argue,
sets the material baseline from which the inﬂuence of altruism and identity emerges.
Material self-interest: Redistribution Most political economy arguments start
from the assumption that an individual’s position in the income distribution determines
her preferences for redistribution. The most popular version of this approach is the the-
oretical model proposed by Romer (1975) and developed by Meltzer and Richard (1981).
To recapitulate very brieﬂy, the RMR model assumes that the preferences of the median
voter determine government policy and that the median voter seeks to maximize current
income. If there are no deadweight costs to redistribution, all voters with incomes below
the mean maximize their utility by imposing a 100% tax rate. Conversely, all voters with
incomes above the mean prefer a tax rate of zero. The RMR model implies that more
inequality should be associated with more redistribution. The consensus in the compara-
tive literature on this topic, however, is either that there is no association between market
income inequality and redistribution or, contrary to the prediction of the RMR model,
less market inequality is associated with more redistribution (e.g., Lindert 1996, Moene
and Wallerstein 2001, Alesina and Glaeser 2004, and Iversen and Soskice 2009).
These ﬁndings must be considered with a degree of caution. This is because most
of this literature relies on macro-comparative empirical analyses (with redistribution as
the dependent variable) and does not pay much attention to individual preferences.1
When looking at individual data, in fact, there is some support for the argument that
relative income inﬂuences preferences. Using comparative data, a relative income eﬀect is
found in, among others, Bean and Papadakis (1998), Finseraas (2009), and Shayo (2009).
Using American data, Gilens (2005), McCarty, Poole and Rosenthal (2008), and Page
and Jacobs (2009) (again, among others) ﬁnd similar eﬀects.
1Even the macro-comparative is less unambiguous that the consensus in the literature suggests. Mi-
lanovic (2000) and Kenworthy and Pontusson (2005) show that rising inequality tends to be consistently
associated with more redistribution within countries.
Material self-interest: Insurance There is a considerable literature arguing that
preferences for redistribution result from a desire on the part of some elements of the
population to be insured against a fall in wealth or income (see, for example, Moene
and Wallerstein 2001). These arguments are based on the idea that the inﬂuence of
material self-interest on redistribution preferences should not be limited to a measure of
present income. In the words of Alesina and Giuliano, “(e)conomists traditionally assume
that individuals have preferences deﬁned over their lifetime consumption (income) and
maximize their utility under a set of constraints. The same principle applies to preferences
for redistribution. It follows that maximization of utility from consumption and leisure
and some aggregation of individual preferences determines the equilibrium level of taxes
and transfers” 2011, 1.
Because of the potential to deﬁne material self-interest inter-temporally (as lifetime
consumption/income), this approach opens the door to arguments about the more direct
eﬀects of contemporary relative income (as in Romer 1975 and Meltzer and Richard 1981),
about social insurance and risk (as in Sinn 1995, Moene and Wallerstein 2003, Iversen and
Soskice 2001, Rehm 2009, Mares 2003), and about social mobility and life-cycle proﬁles
(Alesina and Giuliano 2011, Haider and Solon 2006, Benabou and Ok 2001).
All these argument have a common thrust: that the desire for insurance is heteroge-
neous in the population – some individuals are more risk averse, and hence more likely to
demand insurance, while others are more risk neutral or risk seeking and hence less likely
to demand insurance. They are, however, based on two alternative assumptions. The
ﬁrst is that these concerns are very much associated with self-interest – the implication
here is that individuals who have wealth or higher incomes or are in certain occupations
with volatile unemployment histories are more inclined to prefer redistribution because
they see this as one way to insure against signiﬁcant declines in their ﬁnancial fortunes.
The second is that, alternatively, risk aversion may simply be a personality trait that is
heterogeneous in the population. Risk averse individuals are generally more enthusiastic
about entering the “social” risk pool with other members of society. And this enthusiasm
for entering the risk pool is not conditioned on economic circumstances – its not simply
the better-oﬀ insuring against hard times.
Moene and Wallerstein (2001, 2003) articulate the material self-interest insurance
approach most forcefully (cf. also Sinn 1995, Iversen and Soskice 2001, Rehm 2009,
Mares 2003). Their model builds on the assumption that demand for insurance rises
with income, holding risk exposure constant, and stands the RMR model on its head as
far as the predicted association between inequality and redistribution is concerned. As
a mean-preserving decline in inequality implies that the income of the median voter is
higher, Moene and Wallerstein expect that countries with a more egalitarian distribution
of income will have more redistributive government.
Arguments about the importance of insurance have also emphasized the importance
of risk in determining redistribution and insurance preferences. In this vein, Rehm (2009)
argues that, while income captures redistribution preferences, occupation characteristics
capture risk exposure and insurance motivations.2
In a highly inﬂuential article, Iversen and Soskice (2001) argue that exposure to risk
is inversely related to the portability of individual skills. Another strand of this literature
conceives of risk as mobility and hypothesizes that individuals with good prospects for
upward mobility might be less inclined to support redistribution that their present income
would lead us to expect (e.g., Piketty 1995, Benabou and Ok 2001). By the same logic,
we might expect individuals who anticipate downward mobility to be more supportive of
redistribution than their present income would lead us to expect.
2Religiosity may work as an intermediary between individual insurance considerations and demand
for redistribution. Scheve and Stasavage (2006) hypothesize that the psychic beneﬁts of religiosity
can serve (like social spending) to insure individuals against the eﬀects of adverse life events. As a
consequence, religious individuals on average would prefer lower levels of social insurance provision than
Altruism The possibility that other-regarding concerns inﬂuence redistribution prefer-
ences has received increasing amounts of attention in the recent political economy liter-
ature. There is neural evidence that individuals have a dislike for unequal distributions,
independent from social image or potential reciprocity motivations. Tricomi et al. (2010)
use functional magnetic resonance imaging to test directly for the presence of inequality-
averse social preferences in the human brain. In laboratory experiments, individuals have
been shown to have concerns for the welfare of others (see, for example, Charness and
Rabin 2002 and Fehr and G¨achter 2000). A number of alternative models have been
presented to analyze diﬀerent kinds of other-regarding concerns (for reviews, see Fehr
and Schmidt 2006 and DellaVigna 2009). As we will document below, support for re-
distribution exhibits a high degree of heterogeneity in the UK and extends into income
groups whose support for redistribution could not possibly be motivated by short-term
income maximization. Altruism constitutes one plausible reason why aﬄuent individuals
might support redistribution even though its eﬀect is to reduce their disposable income
and their share of total income.
As argued by Alesina and Giuliano, however, “standard neoclassical general equilib-
rium theory can accommodate altruism, i.e., a situation in which one agent cares also
about the utility of somebody else. But altruism is not an unpredictable ’social noise’
to be randomly sprinkled over individuals” (2011, 94). Altruistic concerns need to be
systematized into predictable hypotheses.
In the empirical part of this paper, we make a distinction between altruism as per-
sonality trait (which we emphasize) and the direct eﬀects of inequality on redistribution
preferences. Other regarding preferences are often operationalized as a welfare function
(the willingness of individuals to make sacriﬁces in order to realize welfare gains for those
in society who are worse oﬀ). The kind of altruism we are interested in, therefore, is
slightly diﬀerent from positive inequity aversion. Fehr and Schmidt argue that an “indi-
vidual is inequity averse if, in addition to his material self-interest, his utility increases if
the allocation of material payoﬀs [in his society] becomes more equitable” (2006, 620).3
In this paper, we don’t analyze the eﬀects of macro-inequality (i.e., the welfare function)
on individual preferences. We argue that altruistic individuals are more likely to support
Identity and in-group altruism We also build on a signiﬁcant recent literature ex-
ploring the role of identity on the formation of preferences for redistribution. There
are material self-interest reasons why identity could matter to redistribution preferences.
Group homogeneity could promote information sharing, the identiﬁcation of free rid-
ers, and communication.4In this paper, however, we emphasize the connection between
altruism and group homogeneity. Much of the literature on altruism emphasizes that
other-regarding considerations are bounded by racial, ethnic or religious cleavages or, in
other words, take the form of “in-group solidarity” or “parochial altruism.”5Habyari-
mana et al. aptly summarize this line of argument by recognizing that individuals may
attach positive utility to the welfare of fellow ethnic [but also national or religious] group
members but no utility (or negative utility) to the welfare of non-group members (2007,
There is a clear relationship between this identity approach and the altruism argu-
ments analyzed in the previous section. While being altruistic implies that an individual’s
utility will increases as the poor beneﬁt from more redistribution, identity arguments em-
phasize that this may be dependent on who the poor are. Perceiving the poor as diﬀerent,
these arguments suggest, diminishes the eﬀects of altruism on redistribution. There can
be little doubt that racism has served as an obstacle to redistributive politics in the Amer-
3For an analysis of the diﬀerent models of altruism and their implications, see Dimick, Rueda and
4For an analysis of the mechanisms underlying these eﬀects, see Habyarimana et al. (2007).
5For an analysis of parochial altruism, see Bernhard, Fischbacher and Fehr (2006).
ican case (Luttmer 2001 and Gilens 2009). Alesina and Glaeser (2004) argue persuasively
that the US is not an exceptional case in this respect.6
Identity and Insurance As we pointed out above there are two conjectures regarding
insurance and preferences for redistribution. In one view risk aversion shapes redistri-
bution preferences because individuals, typically those who are better-oﬀ, are concerned
about falling into ﬁnancial hard times. Favouring redistribution is for the most part
simply an expression of material self-interest. This argument is nicely summarized by
Coate and Ravallion. They explain that in suitably risky environments, current generos-
ity may not reﬂect altruism but expected future reciprocity (1993, 2). In this argument,
group heterogeneity is a proxy for a particular risk pool. In the words of Keely and Tan,
identity provides information about an individuals future economic circumstances in an
environment with uncertainty (2008, 945). Identity groups are identiﬁed with particu-
lar proﬁles regarding risk, mobility, etc (as in Piketty 1995 or Benabou and Ok 2001).
Consequently, where the poor are diﬀerent from the general population, members of the
majority identity group woud feel less vulnerable to risk.
While the relationship outlined above is related to insurance as self-interest (it depends
on the perceived likelihood of needing beneﬁts in the future), risk-aversion as a personality
trait should also be aﬀected by identity. In this case, individuals who are risk-averse would
generally be more enthusiastic about entering the social risk pool if other members in
the pool look like themselves. Because insurance concerns are not simply motivated by
material self-interest, they are more likely to be sensitive to non-material factors such as
the identify of other members in the risk pool.
6The crucial variable of interest in this literature is not fractionalization per se, but rather the con-
centration of minorities among the poor. In Alesina and Glaesers words, the focus is on whether “there
are signiﬁcant numbers of minorities among the poor,” in which case “the majority population can be
roused against transferring money to people who are diﬀerent from themselves” (2004, 134).
Summary There are four broad themes in the literature on the political economy of
redistributive preferences. The ﬁrst two focus on the role of material self-interest (both
as redistribution and as insurance); a third highlights the importance of other-regarding
preferences; and a fourth suggests that identity shapes redistributive preferences. The
contribution of this essay is to evaluate the importance of these diﬀerence factors by
deploying rigorous measures of both the dependent and independent variables associated
with these conjectures. We now turn to the empirics.
Our conjecture is that we can leverage improved measures of these diﬀerent individual-
level characteristics in order to better estimate the the micro-foundations of redistributive
3.1 The dependent variable: redistributive preferences
Eﬀorts to measure individual attitudes about redistribution and redistributive poli-
cies directly have not, for the most part, produced results consistent with the classical
model and hence the puzzle persists. This is particularly the case with respect to voters’
preferences regarding redistributive policies. There might be heterogeneity in abstract
preferences for redistribution – such as the poor think that rich should pay more. Based
on the European Social Survey, (Leon 2012), for example, ﬁnds considerable agreement
with statements calling for more redistribution and there is the kind of heterogeneity in
attitudes we expect – the poor are more demanding of such redistribution and the rich
less so. But with respect to preferences for speciﬁc redistributive policies there appears to
be a distinct absence of heterogeneity – often the poor and the rich have similar expressed
preferences. Or the poor are more resistant – see Kuziemko et al. (2011) who speculate
that this resistance might result from “last place aversion”.
Confounding factors likely play a role here. In particular, individuals might simply
be poorly informed (for a range of reasons) about the distribution of income and their
relatively location in the income distribution. A number of recent studies document the
discrepancies between the objective state of income inequality and individual perceptions
of the distribution of incomes. Norton and Ariely (2011) demonstrate that for the U.S.
case there are signiﬁcant discrepancies between actual and perceived levels of inequality.
Since much of the evidence is based on survey data it is diﬃcult to establish the impor-
tance of these confounding factors in explaining redistribution preferences (Alesina and
Giuliano 2009). Cruces, Perez-Truglia and Tetaz (2013) attempt to address this prob-
lem by embedding experimental treatments within a household survey and, while there
is some evidence that exaggerating ones income distribution rank depresses support for
redistributive measures, the eﬀects are weak.
The Tax Rate Preference Vignettes Our intuition is that these opinions about
redistribution that are measured in public opinion surveys are not particularly good
measures of underlying preferences for redistribution. In fact there are a vareity of mea-
surement issues: The questions do not incorporate any notion of a budget constraint;
there are issues related to social desirability; etc. Rather than asking individuals for
their opinions regarding redistribution, we propose a measure that speciﬁcally measures
preferences for redistributive taxation. The measure simply asks individuals how much
diﬀerent income earners should be taxed. Respondents are simply provided with a set
of scenarios describing a household and its total earnings and are asked how much they
should pay in taxes.
The scenarios varied on three factors. First, hypothetical taxpayers were one of three
types: a single person; a married couple, one of whom worked (outside the home); or
a married couple, both of whom worked. Each respondent was presented with all three
hypothetical taxpayer scenarios involving a single taxpayer, plus two scenarios involving
two-wage couples, and one scenario featuring a one-wage couple. Second, incomes were
set to range in the interval between £8,000 and £80,000. For a two-adult household
without children in the U.K., the median gross income for the top decile is £88,500;
£58,300 for the ninth decile; £46,400 for the eigth; £37,900 for the seventh; £32,200 for
the sixth; £27,200 for the ﬁfth; £23,000 for the forth; £19,900 for the third; £17,100 for
the second; and £13,300 for the bottom.
Several features of the design deserve comment. First, we avoid vague categories such
as “the rich” or “poor people” which respondents might interpret diﬀerently. These hypo-
thetical taxpayers were described as having speciﬁc annual incomes, to avoid ambiguities
and relative comparisons which would make responses incomparable when diﬀerent re-
spondents have diﬀerent ideas about what it means to be “rich” or “poor.” Moreover,
our design ensured that we obtained a great deal of useful variance n taxpayer income,
especially compared to such crude contrasts as “rich” versus “poor” or “middle income.”
Second, we diﬀerentiate clearly between diﬀerent types of taxpayer. The income tax
has, since its inception, treated families diﬀerently from individuals, though few survey
items about taxes take this into account. Our hypothetical taxpayers are all clearly
described as being either single or married. We also tell the respondent how many
dependents the taxpayer has. Dependents, of course, ﬁgure prominently into the current
Question wording. Respondents to the CCAP UK online survey were asked the fol-
lowing worded question regarding the scenarios: “We are going to describe some tax
payers: one tax payer is a single person; another is a married couple ﬁling a joint return.
These tax payers will have diﬀerent income levels; they will also have diﬀerent numbers
of children who are dependent on them; and they will also diﬀer in terms of their their
marital status. In the boxes provided would you indicate the total income taxes you
think each tax payer should pay?”
Distribution of Preferences for Redistributive Tax Rates The marginal tax rate
vignettes provide a unique perspective on precisely what tax rates the average voter con-
siders to be appropriate for diﬀerent types of tax payers. In general there results suggest
that tax preferences respond in a “progressive” fashion – rising with income and declining
with family size. Figure 1 presents summary views of the rates of taxation preferred
by the British and German publics and variations in these rates. Three conclusions can
be drawn from these results: 1) the average rises from about 6 percent to 15 percent
depending on the income/family proﬁle; 2) the range in preferred tax rates around each
of these point estimates is relatively low; and 3) there is very little support for tax rates
that rise above 20 percent.
#8,000## #12,000## #20,000## #40,000## #80,000##
Figure 1: Tax Rate Preferences for Tax Payers in Diﬀerent Income Categories: UK 2009
Figure 2 provides a summary of revealed redistributive tax preferences. First note
that there is never less than 40 percent of respondents who express a preference for a
very low tax rate of 10 percent or less. For the three lowest income brackets about 60
percent choose this 10 percent tax rate while for the higher two income brackets (£40,000
and £80,000) the percentage is closer to 40. And there is an equal amont of unenthusiasm
for high rates of taxation. Regardless of income level, there is never more than 15 percent
of respondents expressing a preference for tax rates of 30 percent or greater. And with
the exception of the largest income bracket (£80,000), less than 5 percent of respondents
favour tax rates of 30 percent or greater. The lowest three income brackets would roughly
fall in the 20 percent tax regime in the U.K. – most respondents seem to think they should
pay about half this amount. For the two highest income brackets, the U.K. tax rate is
40 percent. Only a very small percent of respondents think that this is an appropriate
tax rate – the overwhelming majority of respondents think the tax rate on these income
earners should be half this rate if not lower.
7%# 8%# 8%#
4%# 3%# 3%#
£8,000# £12,000# £20,000# £40,000# £80,000#
Figure 2: Average Prefered Tax Rates by Taxpayer Income Bracket
The descriptive results summarised in Figure 2 oﬀer an important insight into public
preferences for redistributive taxation – on the whole most individuals do not seem to
want too much of it. The modal preferred tax rate across all income brackets is 10 percent
(for most tax brackets a majority of respondents) and there is very little enthusiasm for
tax rates that exceed 30 percent. When one employs unobtrusive measures of preferences
for redistributive taxation, the results suggest the public prefers relatively low tax rates
with little variation over income brackets. This of course is a relatively neutral framing
of redistributive taxation – for example, there is no priming regarding the uses to which
these tax revenues would be put or partisan priming of taxation.7All we have here is a
very neutral recovery of underlying preferences for redistributive taxation.
Nevertheless, in spite of this neutral framing there is variation in preferred tax rates
across income brackets. Figure 2 makes clear that particularly in the case of the highest
income bracket (£80,000), respondents are less likely to prefer low tax rates and con-
siderably more respondents favor either a 20 percent, or 30 percent or greater, rate of
taxation. The principal interest of this essay is understanding who is more, or less, likely
to favor the more redistributive tax rates.
3.2 Material Self-Interest
The tax vignettes suggested that the UK population prefers relatively low levels of
taxation; certainly compared to actual tax rates in the country. Nevertheless, and as we
expected, there is considerable heterogeneity in this regard within the UK population.
The classic conjecture of course is that this variation will be associated with material self-
interest – the rich will exhibit preferences for lower taxation of higher income brackets
while the poor will prefer exactly the opposite.
We need to exercise prudence in exploring the relationship between respondents’ in-
7See for example Lamberton, Neve and Norton (2014) who demonstrate the sensitivity of tax com-
pliancy to agency over the deployment of tax revenues.
come and tax preferences because the income distribution of respondents in the sample
generated from the YouGov internet panel does not reﬂect the actual distribution of in-
come in the UK population. Figure 3 compares the two distributions As you can see, the
CCAP sample signiﬁcantly underrepresents those earning less than £15,000. And the
CCAP sample overrepresents those in the higher earning brackets – those earning more
<£15,000$ £15,000/£30,000$ £30,000/£50,000$ £50,000/£100,000$ >£100,000$
Figure 3: Distribution of Single Tax Payer Income from ONS compared to CCAP Income
3.3 Risk Aversion
Our broad conjecture in this essay is that there are distinct personality traits that
are correlated with redistributive preferences and that the population is heterogeneous
with respect to these characteristics. The argument is essentially that individuals who
demonstrate risk averse behaviour in particular domains, such as health, ﬁnance, retire-
ment or recreation, will engage similar risk preferences regarding redistributive taxation.
For example, individuals who engage in extreme sports are more likely to prefer high risk
equities over very safe annuities and are more likely to self insure for medical emergen-
cies. Similarly, we expect these risk-seeking types to exhibit little appreciation for the
insurance consequences of redistributive taxation. By contrast, risk-averse types in the
population are expected to have a taste for insurance and risk-pooling.
The evidence for a general risk preference trait is persuasive and builds on early
work by psychologists Jackson, Hourany and Vidmar (1972) and more recent work in
economics Dohmen et al. (2011). A signiﬁcant literature has developed that focuses on
measuring this risk preference trait and there is an ongoing debate regarding optimatl
measurment strategies Dohmen et al. (2011), Szrek, Chao and Ramlagan (2012), ?. The
measure of risk preferences that we employ consists of asking respondents whether they
had ever participated in eighteen diﬀerent activities that have varying levels of riskiness.
The activites along with the percentage of respondents indicating they had participated
is summarise in Table 1. We employ a composite measure of risk aversion that is simply
the factor scores estimated for each individual based on a factor analysis of all eighteen
items. High scores on this measure indictate respondents are risk seeking. General risk
attitude survey questions of this nature have been shown to be strongly correlated with
actual risk-taking behaviour Dohmen et al. (2006).
Table 1: Frequency of risky activities in the UK sample
Risky activity % Participating
Taking part in football pools 4.2
Taking part in check pools .5
Playing bingo 5.4
Playing poker 3.1
Betting on the horses 10.1
Playing bid whist .1
Shooting dice .1
Buying lottery tickets 54.8
Speculating on land .4
Playing bridge 1.5
Playing the numbers .3
Entering magazine contests 10.3
Playing roulette 1.5
Playing pinoche .1
Taking part in cricket pools .05
Buying sweepstakes tickets 3.2
Buying raﬄe tickets 29
Buying stocks and bonds 14.5
We also argue that individuals can be altruistic and that the range of this altruism
varies signiﬁcantly across individuals in the population. And again we conceptualise
altruism as a personality trait – there are selﬁsh and unselﬁsh types in the population
and these traits are correlated with a range of behaviours. Just as altruistic types are
more likely to give to charity or volunteer for community work, they are also more likely
to have preferences for redistribution and redistributive taxation.
Private other-regarding preferences are deﬁned here as the underlying other-regarding
behaviour that is measured when employing other-regarding experimental games. In
behavioural economics there has emerged a widely-accepted set of experimental strategies
for recovering these preferences. Early experimental work revealed that subjects playing
games in labs would exhibit unselﬁsh behaviour. For example, the ﬁrst lab experimental
results from the dictator game exhibited that substantial numbers of subjects deviated
from the equilibrium prediction of giving nothing to the other player Forsythe et al.
(1994). Similar unselﬁsh behaviour has been documented in the case of the Ultimatum
Game, Prisoners’ Dilemma Game, and Public Goods Games. Hence there is strong
evidence that subjects have other-regarding behaviour and, as Andreoni and Miller (2002)
demonstrate, it is possible to explain altruistic choices with quasi-concave utility functions
for individuals’ altruism. This demonstrates that altruistic choices are rational.
We propose measuring other-regarding preferences using a standard incentive-
compatible Dictator Game embedded in an online survey conducted in the U.K. in 2010.
We use contributions in the Dictator Game as our proxy for other-regarding preferences.
This allows us to have a rigorous measure of other-regarding preferences for a large
national sample; one that reﬂects that standard measurement strategy for students of
8The precise question wording of the embedded Dictator Game is provided in the Appendix.
010 20 30 40
Percent of Respondents
05000 10000 15000 20000 25000
Amount in YouGov Points Given in Dictator Game
Figure 4: UK Contributions in a Dictator Game
4 Explaining Redistributive Tax Rate Preferences
Of particular interest here is the extent to which an individual supports a redistribu-
tive tax rate. Accordingly, we deﬁne support for redistributive tax rates as simply the
ratio of the tax rate preferred for the highest income category to the lowest income cate-
gory in each tax payer type (recall there are four tax payer types as a function of family
size). The mean of these four ratios is the dependent variable in the subsequent models
of tax rate preferences.
4.1 Material self-interest
The classic expectation is that these preferences should be correlated with income.
Table 2 presents the results from the regression of redistributive tax preferences on our
key theoretical variables. As the ﬁrst model indicates, these preferences are not corre-
lated with income. There is very little evidence from this particular sample that income is
correlated with redistributive preferences. But we are reluctant to draw any strong conclu-
sions regarding income and redistribution preferences because of the sample characteris-
tics. As we pointed out earlier, the sample signiﬁcantly underrepresents the lower income
categories in the U.K. population. One could imagine that underrepresentation of this
segment of the income distribution could signiﬁcantly weaken the income-redistribution
Our measure of risk aversion in Model 1 in Table 2 is, as expected, negative and sta-
tistically signiﬁcant. Risk attitudes are signiﬁcantly related to redistributive preferences.
The correlation is consistent with claims that those who are risk seeking (those who score
high on the risk aversion measure) are less enthusiastic about redistribution while those
Table 2: OLS Regression: High/Low Tax Ratio: UK 2010
Variable Total Low Ethnic High Ethnic
Ethnicity −0.59 −1.20 −0.07
(0.35) (0.52) (0.49)
Income 0.03 0.06 −0.01
(0.04) (0.04) (0.06)
Other-regarding 0.57 0.63 0.33
(0.21) (0.26) (0.37)
Risk Aversion −0.35 −0.54 −0.02
(0.14) (0.17) (0.23)
Constant 1.39 1.21 1.75
(0.40) (0.49) (0.70)
Obs 736 462 274
R-Square 0.02 0.05 0.00
who are risk averse (those scoring low) embrace redistribution. Its not entirely obvious
from the results in Table 2 that, as some claim (Beckman (2006), risk aversion is corre-
lated with ones economic circumstances – speciﬁcally that the rich are risk averse given
their income endowment – while the poor are more risk seeking in the anticipation of
become wealthier. Income is clearly not directly related to redistributive tax preferences.
There is some suggestive evidence that the risk aversion correlations in the models in
Table 2 reﬂect the hypothesised eﬀects of one’s economic circumstances. The suggestive
evidence is that risk preferences are weakly correlated with income – for the bi-variate
regression of risk aversion on income, the income coeﬃcient is -0.2 with a standard error
of .008 and an intercept of .12 (standard error of 0.07).
A more plausible interpretation of the correlation in Table 2 is that our risk aver-
sion measure is capturing an underlying personality trait that is heterogeneous in the
population. There are clearly types in the population who, regardless of their material
circumstances, have a taste for insurance, i.e., are more risk averse.
As we pointed out earlier, altruism may be an important factor determining citizens’
preferences for redistributive tax rates (see Fong (2001) on altruism and redistributive
tax preferences). We contend that altruistic preferences are essentially personality traits
that are not systematically related to income. Our expectation is that altruism will
be distributed similarly amongst the poor and the rich. There are those who contend
otherwise Hoﬀman (2011). The data in fact are consistent with our conjecture that
income is not systematically related to altruism – the correlation between income and
generosity in the Dictator game is essentially zero.9
Figure 5 suggests that in fact this may be the case. The bar representing the total
sample indicates that the most other regarding citizens are more likely to support redis-
tributive tax rates. Figure 5 also compares the rich and poor within each of the other-
regarding categories. Here it is quite clear that the other-regarding eﬀect is conﬁned to
the poor. The rich essentially are unenthusiastic about redistribution regardless of their
other-regarding preferences. On the other hand, its the other-regarding poor who most
strongly support redistributive taxation. Non-other-regarding poor respondents have a
low mean score on the redistribution ratio.
Altruism is, as we hypothesized, correlated with redistributive preferences. Return-
ing to Model 1 in Table 2 we see that the coeﬃcient on altruism is positive as expected
and statistically signiﬁcant. The fact that altruism is quite strongly correlated with re-
distributive preferences, while being essentially orthogonal to income, lends considerable
credence to our contention that personality traits play an important role in explaining
redistributive preferences. Altruism is heterogeneous in the population and is uncorre-
9Its important to point out here that income refers to the actual income of respondents. There is
also evidence from lab experiments that subject endowments – which are frequently treated as “income”
– can have an eﬀect on generosity towards recipients. So for example the distribution of endowments
between recipients and givers can have an eﬀect on generosity of giving Engel (2011), Korenok, Millner
and Razzolini (2012), Branas-Garza (2006)
Selﬁsh" Middle" Generous"
Figure 5: Other-regarding Preferences and Tax Rate Preferences
lated with material self-interest and hence is a critical variable in explaining individual
variation in redistributive preferences.
4.4 Ethnic Heterogeneity
Identity arguments suggest that the the utility that altruistic individuals receive from
the redistributive beneﬁts going to the poor may be dependent on who the poor are. If
the poor are perceived as diﬀerent this diminishes the eﬀects of altruism on redistributive
preferences. These identify argments also have implications for how risk aversion shapes
redistributive preferences. The identify conjecture suggests that risk-averse individuals
are more likely to be enthusiastic about entering the social risk pool if other members in
the pool look like themselves. We can explore these conjectures with the BCCAP survey
The fourth through sixth waves of the BCCAP survey included a measure of ethnicity.
In order to measure ethnic heterogeneity we used this question to created a variable that
assumed a value of 0 if the respondent answered white and a 1 for all other responses.10
The argument concerns context so we summarised the mean value of this variable at the
government oﬃce of the region level (there are 11 of these regions).
Model 2 and Model 3 of Table 2 present results for the estimation of the model for
respondents in ethnically diverse contexts and those in ethnically homogeneous contexts.
According to the ﬁrst conjecture, levels of ethnic heterogeneity might suppress the cor-
relation between altruism and support for redistributive tax rates. Model 2 and Model
3 suggest that this very well may be the case. The coeﬃcient on altruism is relatively
small in ethnically diverse contexts (0.33) and has a large standard error compared to
the coeﬃcient in ethnically homogeneous contexts that is relatively high (0.63) and is
It has become widely accepted that support for redistribution is probably not nar-
rowly driven by simple material self-interest. This essay provides some further insight
into factors, other than material self-interest, that might shape one’s preferences for
redistribution. We focus speciﬁcally on support for redistributive taxation. And we im-
plement a rigorous strategy for recovering, in a relatively unobtrusive fashion, individuals’
preferences for redistributive taxation. The survey data is from an online panel survey
10The ethnic identify answer categories were the following: white British; any other white background;
white and black caribbean; white and black african; white and asian; white and asian; any other mixed
background; indian; pakistani; bangladeshi; any other asian background; black caribbean; black african;
any other black background; chinese; other ethnic group. The ﬁrst two white answer categories were
coded as white - the rest were coded as other ethnic.
conducted as part of the 2008-2010 UK BCCAP project. An important caveat here is
that the sample tends to underestimate lower income categories in the U.K. population.
With that caveat in mind our data suggest that individuals support a fairly narrow range
of tax rates that are not particularly high. Depending on the nature of the household
and income levels, our respondents support tax rates that ranged between about 6 and
But our interest here is primarily understanding variation in preferences for redis-
tributive tax rates. We ﬁnd that income is essentially uncorrelated with redistributive
preferences. And while this result is consistent with other individual-level studies our
sample characteristics lead us to be relatively cautious about the ﬁnding. Rather, we
are particularly interested in the role that individual-level traits play in forming one’s
preferences for redistribution. We explore whether altruistic and risk-averse personality
types are more likely to support redistributive taxation. We distinguish the eﬀects of an
altruistic personality on redistributive preferences from those of positive inequity aversion
that derives from an analysis of macro-inequality in an individul’s utility function. Simi-
larly, we argue that it is important to distinguish the eﬀect on redistribution preferences
of risk-averse personality from the eﬀect of risk aversion (or insurance) that is the result
of materialist self-interest considerations. An example of the latter is a concern that the
unemployment history of one’s occupation makes individuals insure against signiﬁcant
declines in their ﬁnancial fortunes. Our results clearly suggest that it is the altruistic and
risk-averse personality trait that accounts for a considerable amount of the heterogeneity
in redistributive preferences.
Finally, we argue that the ethnic diversity of the context in which personality types
ﬁnd themselves will condition the intrinsic utility they derive from altruism or insurance.
Our results suggests that the utility that altruistic individuals receive from the redis-
tributive beneﬁts decline if the poor are perceived as diﬀerent this diminishes the eﬀects
of altruism on redistributive preferences. The ﬁndings also indicate that risk-averse in-
dividuals are more likely to be enthusiastic about entering the social risk pool if other
members in the pool look like themselves.
6.1 Measuring Redistributive Tax Preferences
Types of Taxpayers. The following tables indicate the types of taxpayers for which
respondents need to assign a total income tax liability.
Table 3: Income and Dependents Table: Single Person
8,000 12,000 20,000 40,000 80,000
Table 4: Income and Dependents Table: Married Couple – One Working Outside Home
8,000 12,000 20,000 40,000 80,000
Table 5: Income and Dependents Table: Married Couple Both Working Outside Home
8,000 12,000 20,000 40,000 80,000
6.2 Altruism Measure: U.K.
The measure of altruism for the U.K. is based on the administration of a simple
dictator game in an online survey conducted in 2010. The question was worded as follows:
UK: In addition to the normal SSI points cash prize (30 points) for completing the
survey, we will be drawing one other prize at the end of this survey. As you know people
across the country are being interviewed for this survey. After all of the interviews are
completed, two of the people interviewed will share a prize of 40000 SSI points (£400).
Here is how the winners of the money will be decided:
Each person interviewed will indicate how much of the 40000 Points they want to give
to themselves. After we have ﬁnished all of our interviewing, we will randomly select one
of the participants in this survey. If you are selected, you will be paid whatever amount
of the 40000 SSI points you decided to give to yourself. Then another person’s name will
be randomly selected. This second person will receive whatever amount remains of the
40000 points. Your decision will determine how much money you might receive so please
take the decision seriously.
In the event that your name is chosen ﬁrst, we want you to decide right now how
much of the 40000 points prize you want to give to yourself. Please choose the amount
you want to give yourself, and the amount you want to leave for someone else.
Alesina, A. and P. Giuliano. 2009. “Preferences for Redistribution.” NBER Working
Alesina, Alberto and Edward L Glaeser. 2004. Fighting Poverty in the US and Europe.
A world of Diﬀerence. Oxford: Oxford University Press.
Alesina, Alberto and Paola Giuliano. 2011. Preferences for Redistribution. In Handbook
of Social Economics, ed. Jess Benhabib, Alberto Bisin and Matthew O. Jackson. San
Diego: North-Holland pp. 93–131.
Andreoni, James and J. Miller. 2002. “Giving According to GARP: An Experimental
Test of the Consistency of Preferences for Altruism.” Econometrica 70:737–53.
Bean, Clive and Elim Papadakis. 1998. “A Comparison Of Mass Attitudes Towards The
Welfare State In Diﬀerent Institutional Regimes, 1985–1990.” International Journal of
Public Opinion Research 10(3):211–236.
Beckman, Steven R. 2006. “A Tax and Redistribution Experiment with Subjects that
Switch from Risk Aversion to Risk Preference.” Social Choice Welfare 27:627–641.
Benabou, R. and E. A. Ok. 2001. “Social Mobility and the Demand for Redistribution:
The Poum Hypothesis.” The Quarterly Journal of Economics 116(2):447–487.
Bernhard, Helen, Urs Fischbacher and Ernst Fehr. 2006. “Parochial altruism in humans.”
Branas-Garza, P. 2006. “Poverty in Dictator Games. Awakening Solidarity.” Journal of
Economic Behavior and Organization 60:306–320.
Charness, Gary and Matthew Rabin. 2002. “Understanding social preferences with simple
tests.” The Quarterly Journal of Economics 117(3):817–869.
Coate, Stephen and Martin Ravallion. 1993. “Reciprocity without commitment: Charac-
terization and performance of informal insurance arrangements.” Journal of develop-
ment Economics 40(1):1–24.
Cruces, G., R. Perez-Truglia and M. Tetaz. 2013. “Biased Perceptions of Income Dis-
tribution and Preferences for Redistribution: Evidence from a Survey Experiment.”
Journal of Public Economics 98:100–112.
DellaVigna, Stephano. 2009. “Psychology and Economics: Evidence from the Field.”
Journal of Economic Literature 47(2):315–372.
Dimick, Matthew, David Rueda and Daniel Stegmueller. 2014. “Inequality, Fairness, and
Preferences for Redistribution?” Manuscript.
Dohmen, Thomas, Armin Falk, David Hoﬀman, J. Schupp and G. Wagner. 2011. “In-
dividual Risk Attitudes: Measurement, Determinants, and Behavioral Consequences.”
Journal of the European Economic Association 9:522–550.
Dohmen, Thomas, Armin Falk, David Hoﬀman, Jurgen Schupp, Uwe Sunde and
Gert Georg Wagner. 2006. “Individual Risk Attitudes: New Evidence from a Large,
Representative, Experimentally-Validated Survey.” Centre for Economic Policy Re-
Engel, Christoph. 2011. “Dictator Games: A Meta Study.” Experimental Economics
Fehr, Ernst and Klaus M Schmidt. 2006. “The economics of fairness, reciprocity and
altruism–experimental evidence and new theories.” Handbook of the economics of giv-
ing, altruism and reciprocity 1:615–691.
Fehr, Ernst and Simon G¨achter. 2000. “Fairness and retaliation: The economics of
reciprocity.” The journal of economic perspectives pp. 159–181.
Finseraas, Henning. 2009. “Income Inequality and Demand for Redistribution: A Multi-
level Analysis of European Public Opinion.” Scandinavian Political Studies 32(1):94–
Fong, C. 2001. “Social Preferences, Self-Interest, and the Demand for Redistribution.”
Journal of Public Economics 82(2):225–246.
Forsythe, Robert, Joel Horowitz, N.S. Savin and Martin Sefton. 1994. “Fairness in Simple
Bargaining Games.” Economic Behavior 6:347–369.
Gilens, Martin. 2005. “Inequality and Democratic Responsiveness.” Public Opinion Quar-
Gilens, Martin. 2009. Why Americans hate welfare: Race, media, and the politics of
antipoverty policy. University of Chicago Press.
Habyarimana, James, Macartan Humphreys, Daniel N. Posner and Jeremy M. Weinstein.
2007. “Why Does Ethnic Diversity Undermine Public Goods Provision?” American
Political Science Review null:709–725.
Haider, Steven and Gary Solon. 2006. “Life-cycle variation in the association between
current and lifetime earnings.” American Economic Review 96(4):1308–1320.
Hoﬀman, Mitchell. 2011. “Does Higher Income Make You More Altruistic? Evidence
from the Holocust.” The Review of Economics and Statistics 93(3):876–887.
Iversen, Torben and David Soskice. 2001. “An Asset Theory of Social Policy Preferences.”
American Political Science Review 95(4):875–893.
Iversen, Torben and David Soskice. 2009. “Distribution and Redistribution: The Shadow
of the Nineteenth Century.” World Politics 61(03):438.
Jackson, D., L. Hourany and N. Vidmar. 1972. “A Four-dimensional Interpretation of
Risk-taking.” Journal of Personality 40:483–501.
Keely, Louise C and Chih Ming Tan. 2008. “Understanding preferences for income redis-
tribution.” Journal of Public Economics 92(5):944–961.
Kenworthy, Lane and Jonas Pontusson. 2005. “Rising inequality and the politics of
redistribution in aﬄuent countries.” Perspectives on Politics 3(3):449–471.
Korenok, O., E.L. Millner and L. Razzolini. 2012. “Are Dictators Averse to Inequality?”
Journal of Economic Behavior and Organization 82:543–547.
Kuziemko, Ilyana, Ryan W. Buell, Taly Reich and Michael I. Norton. 2011. “‘Last-Place
Aversion’: Evidence and Redistributive Implications.” National Bureau of Economic
Research Working Paper 17234.
Lamberton, Cait, Jan-Emmanuel De Neve and Michael I. Norton. 2014. “Eliciting Tax-
payer Preferences Increases Tax Compliance.” Manuscript.
Leon, Francisco Jose. 2012. “Reciprocity and Public Support for the Redistributive Role
of the State.” Journal of European Sociology 22(2):198–215.
Lindert, Peter H. 1996. “What limits social spending?” Explorations in Economic History
Luttmer, Erzo FP. 2001. “Group loyalty and the taste for redistribution.” Journal of
political Economy 109(3):500–528.
Mares, Isabela. 2003. The Politics of Social Risk. New York: Cambridge University Press.
McCarty, Nolan, Keith T Poole and Howard Rosenthal. 2008. Polarized America: The
Dance of Ideology and Unequal Riches. The MIT Press.
Meltzer, Allan H and Scott F Richard. 1981. “A Rational Theory of the Size of Govern-
ment.” Journal of Political Economy 89(5):914–927.
Milanovic, Branko. 2000. “The median-voter hypothesis, income inequality, and income
redistribution: an empirical test with the required data.” European Journal of Political
Moene, Kark Ove and Michael Wallerstein. 2001. “Inequality, Social Insurance and Re-
distribution.” American Political Science Review 95(4):859–874.
Moene, Karl Ove and Michael Wallerstein. 2003. “Earnings Inequality and Welfare Spend-
ing: A Disaggregated Analysis.” World Politics 55(4):485–516.
Norton, M. and D. Ariely. 2011. “Building a Better America – One Wealth Quintile at a
Time.” Perspectives on Psychological Science 6:9–12.
Page, Benjamin I and Lawrence Jacobs. 2009. Class War? What Americans Really Think
about Economic Inequality. Chicago: University of Chicago Press.
Piketty, T. 1995. “Social Mobility and Redistributive Politics.” Quarterly Journal of
Rehm, Philipp. 2009. “Risks and Redistribution: An Individual-Level Analysis.” Com-
parative Political Studies 42(7):855–881.
Romer, Thomas. 1975. “Individual welfare, majority voting and the properties of a linear
income tax.” Journal of Public Economics 4:163–185.
Scheve, Kenneth and David Stasavage. 2006. “Religion and Preferences for Social Insur-
ance.” Quarterly Journal of Political Science 1(3):255–286.
Shayo, Moses. 2009. “A Model of Social Identity with an Application to Political
Economy: Nation, Class, and Redistribution.” American Political Science Review
Sinn, Hans-Werner. 1995. “A Theory of the Welfare State.” Scandinavian Journal of
Szrek, Helena, Li-Wei Chao and Shandir Ramlagan. 2012. “Predicting (un) Healthy
Behavior: A Comparison of Risk-taking Propensity Measures.” Judgment and Decision
Tricomi, Elizabeth, Antonio Rangel, Colin F Camerer and John P O’Doherty. 2010.
“Neural evidence for inequality-averse social preferences.” Nature 463(7284):1089–1091.